Testing 10 deer samples to see differences within BLAST algorithms (and the accuracy of BLAST itself) for ambiguous ASVs in significant quantities.
BLASTN: slow, but allows word-size down to seven bases.
Within the hovertext there is a line called SpeciesASV. This is what IDTAXA (Th30) has classified the ASV as.
Notice the lighter green (Haemonchus contortus). For this plot, I have left all classifications how they were outputted by BLAST to show how drastic the difference could look. For the next two plots, I have manually “fixed” the classification of Haemonchus contortus based on the manual revision of the ASV on Geneious.
DISCONTIGUOUS MEGABLAST: uses initial seed that ignores some bases (allowing mismatches) - intended for cross-species comparisons.
Notice here “Haemonchus contortus”" is now “Haemonchus contortus but actually placei”. The rest of Haemonchus contortus (ASV 7 and ASV 21) was analyzed and discovered to be classified as such due to poor query coverage (16%), which is reflected in the pipeline output as a low bitscore.
MEGABLAST: comparing a query to closely related sequences - works best if target percent identity >95%.